Nutrition applications like Yuka or Codecheck, just to name the most popular, are decrypting food labels to provide nutritional or other information. Their marketing messages say that they help consumers to spot healthy foods who, once better informed, can then shop in good conscience. But should we put our blind trust in them?

The premise of these apps: nutritional databases

The approach of these nutrition applications (such as Yuka, Codecheck, etc.) is to provide nutritional information when customers scan the barcode of food products. The application then displays information found from food databases.

These applications process the information, make it understandable, and evaluate it according to various criteria defined by their developers. The evaluation aspect reassures clients that they’re making the right purchasing decision.

As these applications are used by millions of consumers, though, their services are scrutinized (or even criticized) by their competitors, dietary experts and agri-food brands whose products are then subject to this analysis. We also have an opinion on the matter.

Unreliable source of nutritional data

One main point of contention about these nutrition applications is the product assessment. What are the evaluation criteria? Are they objective? For example, the presence of additives such as E102 (a dye) and E234 (a preservative) is hotly debated for being potentially harmful. This subject is divisive and often leads to standoffs between those “for” and those “against.”

Aside from the subjective nature of certain food product classifications, the second criticism concerns the accuracy and qualification of information sources.

Some nutrition applications, for example, use the OpenFoodFacts food product database. While deserving of merit for their independence, these databases are open and collaborative. Essentially, consumers voluntarily feed it data. It’s kind of like the Wikipedia of food.

But this database is not verified by professionals. Even so, a considerable number of errors are introduced directly at the industrial level, during nutritional declaration. So even labels aren’t completely reliable.

Can nutrition applications guarantee good food decisions?

A polluted database is much like a polluted source. If you produce water that’s bottled from a polluted source, the rest of the supply chain is then tainted. You may have the most beautiful bottles, amazing labels and great sellers, but you’ll always have unclean water. And if you use this water while cooking, everything it’s in will be just as polluted.

No matter how beautifully presented, the information you get on your smartphone will have errors if your application is based on an unverified source of information. So, that means there’s no guarantee that you’ve purchased healthy food.

Les applications nutritionnelles, garantes de nos bons choix alimentaires ?

A high error rate in nutritional information

What are our findings? Our last case study was very revealing. We verified a brand’s 5,000 food products. Our analysis focused primarily on the presence of allergens and the levels of fat, calories, salt, sugar, etc. To our surprise, we found a significant error rate.

No less than 66% of the analyzed products have been subject to correction by Youmeal. Among the products whose information was incorrect, allergen declarations were incomplete or missing 19% of the time and, for 14% of cases, there was incorrect macronutritional information.

Imagine you mix up some ingredients, for which the information is already incorrect, while making a recipe. The number of errors then dramatically increases. Good luck giving the correct nutritional information for your recipe! Nutrition applications are of no help in this case.

Getting reliable information: a matter of time and money

The most commonly used method for obtaining reliable and verified information is calling a dietitian, who: takes the time to read labels one-by-one, connects the list of ingredients with macronutritional and allergenic data and can detect errors in nutritional databases. And then propose corrections.

Definitely not a fun task and one that could introduce more errors due to its repetitive nature. Not to mention how time-consuming such a study of thousands of products would be, which is a major obstacle. As they say, time is money.

So let’s take a look at the numbers. Manually analyzing and correcting 5,000 nutritional declarations would mean verifying approximately 60,000 numbers. At the rate of one minute per figure, that’s more than 6 months of work (and payment). Which is exactly why the verification of collaborative food databases is practically nonexistent. Who would be willing or able to pay for this? And let’s not forget that there are hundreds of thousands of food items out there on the market, not counting the ~10% that are added each year.

How technology supports quality nutritional information

Based on these observations, Youmeal developed a food tech software that’s capable of reading and understanding the ingredients list (even in several languages, with spelling mistakes and abbreviations). Unlike existing nutrition applications, the software is based on advanced algorithms that read, understand, verify, and correct nutritional information in record time. This field is complex, though, so human intervention is still needed to resolve ambiguities. But algorithms are already doing most of the work.

For example, thanks to its algorithms, Youmeal took less than a month to analyze and verify our client’s 5,000 products… that’s six times faster than a human! We think that the technology will be ten times faster in just a few months. How does it work? Algorithms find the connection between the nutritional data available on labelling and that which is found in the Youmeal database. In the end, this artificial intelligence is constantly learning and delivers very precise information.

Managers of commercial or private kitchens can be certain to receive accurate nutritional information on all of their recipe ingredients when they use our software. Producers and distributors of these products (the wholesalers) will be able to sleep soundly knowing that Youmeal has analyzed their nutritional declarations.

Interested in a more detailed presentation of our case study, how Youmeal solves these problems or what solutions we offer? Attend our Webinar on 21 March at 3pm. We’ll discuss the nutrition declaration: the approaches and tools to protect ourselves from the current lack of accuracy. Registration is free!